Meaning
Statistical outlier detection limits utilize a multi-dimensional metric to measure the distance of a point from the distribution center of a benchmark dataset. Quality engineers apply a mahalanobis distance threshold to identify anomalous batches of plastic resin during raw material screening. This analytical boundary operates exclusively in multi-variable space and cannot be applied to isolated, single-variable measurements.
Mathematical Calculation
Covariance matrices of the baseline resin samples are generated to account for the correlations between different spectral or physical properties. The mahalanobis distance threshold is calculated from these matrices to establish the maximum acceptable variance for incoming shipments. When a new lot is scanned, its calculated distance is compared against this critical value to determine if the sample belongs to the historical population of conforming materials.
This approach evaluates the entire dataset simultaneously, preventing individual variations from masking a compound anomaly.
Calibration Procedure
Reference libraries of conforming resin lots must be compiled over several production runs to establish a stable baseline. The mahalanobis distance threshold must be adjusted whenever there is a planned change in the raw material formulation or the polymerization reactor. This calibration ensures that normal process variation is not misidentified as a quality defect by the analysis software.
Industrial Action
Batches that exceed the established limit are flagged for rejection or further chemical analysis. The mahalanobis distance threshold provides an automated decision-making tool on the factory floor, allowing operators to segregate non-conforming lots before they are introduced into the compounding extruder. This rapid screening prevents the processing of contaminated material that would otherwise cause defects in the finished product.
